| Current research aims towards developing new characterization methodologies of nanostructures to estimate and model nanomaterial behavior using artificial intelligence. The proposed methodology in this research utilizes artificial neural networks and image processing techniques to study the structural and mechanical properties of micro- and nano-structures and to develop a quality estimation approach of these materials. For their engineering significance, carbon nanotubes (CNTs) are the main focus of this research. The proposed scalable process will dramatically improve materials design, the use of these materials in nanotechnology and MEMS, and it will facilitate full scale production. In this research project, a new approach based on artificial neural networks modeling was developed to model the aging behavior of an Al-Mg-Si alloy and to distinguish the precipitate morphology at each stage of aging of this microstructure. An image analysis algorithm capable of capturing orientation gradient, nearest neighbor distances, number density, shapes, and size of precipitates was developed. The neural networks model combines the most important precipitate parameters including volume fraction, shape, size and distance between precipitates extracted by the image analysis. It was found that the model is able to successfully predict the age hardening behavior of AA6022 in both deformed and undeformed conditions. To characterize carbon nanotube samples, we have identified a set of intermediate steps that will lead to a comprehensive, scalable set of procedures for analyzing nanotubes. Image analysis techniques were employed and stereological relations were determined for SEM images of CNT structures; these results were utilized to estimate the morphology of the turf (i.e. CNTs alignment and curvature) using artificial neural networks classifier. This model was also used to investigate the link between Raman spectra of CNTs and the quality of the structure morphology, where strong relations were found for the structural effect on the Raman features. We have also proposed a new methodology to investigate the correlation between indentation resistance of multi-wall carbon nanotube turfs, Raman spectra and the geometrical properties of the turf structure using adaptive neuro-fuzzy phenomenological modeling. This methodology yields a novel approach for modeling at the nanoscale by evaluating the effect of structural morphologies on nanomaterial properties using Raman Spectroscopy. A parametrical study of these features was conducted using artificial intelligence to determine the effectiveness of the involved parameters included in this study. |
